This invention solves a problem related to using Power MOSFET intrinsic diodes to clamp inductive flyback energy in a half-bridge configuration.
Both N-channel and P-channel Power MOSFET devices contain intrinsic diodes that can be used to clamp inductive flyback energy from an inductive load when used in a half-bridge configuration. This eliminates the need for adding discrete fast recovery diodes to clamp the resulting flyback voltage between the supply voltage and ground. Whether the intrinsic Power MOSFET diodes or discrete diodes are used, the reverse recovery time (Trr) is of concern. Any time a diode is in forward conduction, and the opposing device in the half-bridge is switched "on," some recovery time will be required before recombination occurs and the conducting diode begins to block reverse voltage. This can be illustrated with regard to FIG. 1A through 1C. In FIG. 1A, Power MOSFET Q2 is on and Power MOSFET Q1 is off, thus causing current to flow from positive supply terminal 10 through load 11, through conducting Power MOSFET Q2, to ground. As shown in FIG. 1B, when Power MOSFET Q2 turns off, inductive load 11 generates a current which causes the intrinsic diode of Power MOSFET Q1 to be forward biased, thereby clamping the flyback voltage generated by load 11. As shown on FIG. 1C, Power MOSFET Q2 then turns on in order to cause current flow through load 11. However, during a brief time period after Power MOSFET Q2 turns on while the instrinsic diode Power MOSFET was in the forward biased state, the instrinsic diode of Power MOSFET Q1 is free to conduct current in reverse. This creates a low impedance path through reverse conducting intrinsic diode of Power MOSFET Q1 and conducting Power MOSFET Q2, directly from power supply terminal 10 to ground. This so-called "shoot-through current" through this low impedance path can build to component damaging levels during the recovery time of the intrinsic diode of Power MOSFET Q1 if some method of control is not implemented.
In the simple example shown in FIGS. 1A through 1C, transistor Q1 is placed in the circuit solely for the purpose of providing an intrinsic diode. Similarly, in the example of FIGS. 1A through 1C, transistor Q2 includes an intrinsic diode which never conducts in its forward direction, and thereby does not cause a problem with regard to shoot-through current. However, the circuit of FIG. 1D shows an example where two MOSFET transistors Q1 and Q2 are arranged where both are controlled by control currents .phi. and .phi., respectively. Transistor Q1, when turned on, provides a current path between positive supply voltage V+ to inductive load L and thence to ground. Conversely, when transistor Q2 is conducting, a path is formed from ground through inductive load L, and through MOSFET Q2 to negative supply voltage V-. In the embodiment of FIG. 1D, the intrinsic diodes of both MOSFETs Q1 and Q2 will, depending on the state of operation, be forward biased to conduct inductive current, and thus each intrinsic diode of MOSFETs Q1 and Q2 will, during various operating conditions, provide a potential shoot-through problem during their recovery times T.sub.rr.
Discrete diodes, when used to clamp inductive flyback energy, can be sized to handle considerable current during recovery, and fabricated to have fast recovery times to minimize shoot-through current. Unfortunately, simply paralleling the intrinsic diode of a Power MOSFET with a discrete fast recovery diode, to replace switching MOSFETs Q1 of FIGS. 1A through 1C, or either one or both of switching MOSFETs Q1 and Q2 of FIG. 1D, as shown in FIG. 2A, does not solve the recovery time problems. For example, if the discrete diode does not have a significantly lower forward voltage drop, the intrinsic diode will share the forward current and still require some amount of recovery time. If this occurs, and the discrete diode recovers first, the intrinsic diode is left as vulnerable as without the paralleled, fast recovery diode. To assure reliable bypassing of flyback energy through a discrete diode, the addition of a series diode is required as in FIG. 2B to block any forward conduction current through the intrinsic diode of the Power MOSFET. Unfortunately, adding a series blocking diode increases forward voltage drop and negates the advantages of increased efficiency which would otherwise be obtained using Power MOSFETs with low source to drain resistance (R.sub.DS (on)).
Using the intrinsic diodes of Power MOSFETs eliminates the cost, space, and added voltage loss of adding discrete diodes. When using the intrinsic diodes of Power MOSFETs, care must be taken to protect them from destructive dV.sub.DS /dt rates and excessive current during recovery. However, Trr of the Power MOSFET intrinsic diode is not specified as consistently as discrete fast recovery diode, and Trr specifications differ between manufacturers of otherwise compatible Power MOSFETs.
Traditionally the problem of excess shoot-through current during the recovery time of the intrinsic diode of the Power MOSFET has been addressed by simply reducing the gate drive to the Power MOSFET and slowing the turn-on time of the opposing Power MOSFET. This allows the intrinsic diode additional time to recover before shoot-through current builds to an unacceptable level, and reduces the resulting dV.sub.DS /dt after recovery.
FIG. 3 demonstrates one commonly used method of slowing Power MOSFET turn-on rate without increasing the turn-off time. During turn-on of Power MOSFET Q1, Resistor R1 limits the current available for charging the gate capacitance (Ciss) of Power MOSFET Q1, thereby slowing the turn-on of Power MOSFET Q1. Resistor R1 is commonly paralleled by a diode D1 to maintain minimum driver impedance during turn-off, and to hold Power MOSFET "off" in the static state.
Slowing turn-on of the "opposing" Power MOSFET (i.e. the Power MOSFET opposing the Power MOSFET which has an intrinsic diode which is in forward conduction) limits shoot-through current through the Power MOSFET being turned on during recovery of the conducting diode. While this method achieves the goal of limiting shoot-through current, it does so at the expense of switching losses. Since this method requires turn-on of the opposing Power MOSFET to always be based on "worst case" intrinsic diode recovery time, turn on of the opposing Power MOSFET is needlessly slowed if the intrinsic diode was not in forward conduction immediately prior to turn-on of the opposing Power MOSFET, since in this event the intrinsic diode does not require time to recover. Similarly, during those instances in which the intrinsic diode is in forward conduction immediately prior to turn-on of the opposing Power MOSFET, turn-on of the opposing Power MOSFET is needlessly slowed during the time required for transition of the gate voltage to full "enhancement voltage" after the intrinsic diode recovers.
"MOSPOWER Applications Handbook," published by Silicomix Incorporated of Santa Clara, Calif. 95054, 1984, is relevant reference material and is hereby incorporated by reference.